(15), p. 8 — the robust chance constraint holds iff the worst-case VaR is ≤ 0
ProvedDRJointCC.Individual.eq_15chance-constraintdistributionally-robust-optimizationp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1value-at-risk
Let be the set of probability distributions on with mean and covariance , let , and let be a measurable loss function. Then
where is the worst-case Value-at-Risk (14).
This is the first step of the proof of Theorem 2.2: it reduces the distributionally robust chance constraint to a sign condition on the worst-case VaR.
Formalization Note The paper states (15) inside the proof of Theorem 2.2, for a continuous concave or quadratic loss; its argument uses only measurability of , so the statement is made for every measurable , which is stronger. The worst-case VaR is extended-real valued. The standing assumptions and of §2 are hypotheses.
Preamble
import Mathlib import Definitions.Def_DRJointCC_Individual_Individual open MeasureTheory open scoped ENNReal
Formal statement
namespace DRJointCC.Individual
/-- (15), p. 8: under the standing assumptions, the distributionally robust individual chance
constraint holds iff the worst-case VaR (14) is nonpositive, for every measurable loss `L`. -/
theorem eq_15 {k : ℕ} (μ : Fin k → ℝ) (Sig : Matrix (Fin k) (Fin k) ℝ) (hSig : Sig.PosDef)
(ε : ℝ) (hε : 0 < ε) (hε1 : ε < 1)
(L : (Fin k → ℝ) → ℝ) (hL : Measurable L) :
ENNReal.ofReal (1 - ε) ≤ ⨅ P ∈ ambiguitySet μ Sig, P {ξ | L ξ ≤ 0} ↔
wcVaR ε μ Sig L ≤ 0 := by sorry
end DRJointCC.Individual
Source
Zymler, Kuhn, Rustem, Distributionally Robust Joint Chance Constraints with Second-Order Moment Information, Math. Program. 137 (2013), accepted manuscript of 13 Aug 2011, p. 8, (15) (proof of Theorem 2.2)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.